| 1 |
/* Copyright (C) 2018-2025 Free Software Foundation, Inc. |
| 2 |
|
| 3 |
This file is free software; you can redistribute it and/or modify it |
| 4 |
under the terms of the GNU General Public License as published by the |
| 5 |
Free Software Foundation; either version 3, or (at your option) any |
| 6 |
later version. |
| 7 |
|
| 8 |
This file is distributed in the hope that it will be useful, but |
| 9 |
WITHOUT ANY WARRANTY; without even the implied warranty of |
| 10 |
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 11 |
General Public License for more details. |
| 12 |
|
| 13 |
Under Section 7 of GPL version 3, you are granted additional |
| 14 |
permissions described in the GCC Runtime Library Exception, version |
| 15 |
3.1, as published by the Free Software Foundation. |
| 16 |
|
| 17 |
You should have received a copy of the GNU General Public License and |
| 18 |
a copy of the GCC Runtime Library Exception along with this program; |
| 19 |
see the files COPYING3 and COPYING.RUNTIME respectively. If not, see |
| 20 |
<http://www.gnu.org/licenses/>. */ |
| 21 |
|
| 22 |
/* An executable stack is *not* required for these functions. */ |
| 23 |
#ifdef __linux__ |
| 24 |
.section .note.GNU-stack,"",%progbits |
| 25 |
.previous |
| 26 |
#endif |
| 27 |
|
| 28 |
#ifdef L__mulsi3 |
| 29 |
.balign 4 |
| 30 |
.globl __mulsi3 |
| 31 |
.type __mulsi3, @function |
| 32 |
__mulsi3: |
| 33 |
l.movhi r11, 0 /* initial r */ |
| 34 |
|
| 35 |
/* Given R = X * Y ... */ |
| 36 |
1: l.sfeq r4, r0 /* while (y != 0) */ |
| 37 |
l.bf 2f |
| 38 |
l.andi r5, r4, 1 /* if (y & 1) ... */ |
| 39 |
l.add r12, r11, r3 |
| 40 |
l.sfne r5, r0 |
| 41 |
#if defined(__or1k_cmov__) |
| 42 |
l.cmov r11, r12, r11 /* ... r += x. */ |
| 43 |
l.srli r4, r4, 1 /* y >>= 1 */ |
| 44 |
#else |
| 45 |
l.bnf 3f |
| 46 |
l.srli r4, r4, 1 /* y >>= 1 */ |
| 47 |
l.ori r11, r12, 0 |
| 48 |
3: |
| 49 |
#endif |
| 50 |
l.j 1b |
| 51 |
l.add r3, r3, r3 /* x <<= 1 */ |
| 52 |
|
| 53 |
2: l.jr r9 |
| 54 |
l.nop |
| 55 |
|
| 56 |
.size __mulsi3, . - __mulsi3 |
| 57 |
#endif |
| 58 |
|
| 59 |
#if defined(L__udivsi3) || defined(L__umodsi3) \ |
| 60 |
|| defined(L__divsi3) || defined(L__modsi3) |
| 61 |
.global __udivmodsi3_internal |
| 62 |
.hidden __udivmodsi3_internal |
| 63 |
.type __udivmodsi3_internal, @function |
| 64 |
#endif |
| 65 |
|
| 66 |
#ifdef L__udivsi3 |
| 67 |
.balign 4 |
| 68 |
.global __udivsi3 |
| 69 |
.type __udivsi3, @function |
| 70 |
__udivsi3: |
| 71 |
__udivmodsi3_internal: |
| 72 |
/* Note that the other division routines assume that r13 |
| 73 |
is not clobbered by this routine, and use that as to |
| 74 |
save a return address without creating a stack frame. */ |
| 75 |
|
| 76 |
l.sfeq r4, r0 /* division by zero; return 0. */ |
| 77 |
l.ori r11, r0, 0 /* initial quotient */ |
| 78 |
l.bf 9f |
| 79 |
l.ori r12, r3, 0 /* initial remainder */ |
| 80 |
|
| 81 |
/* Given X/Y, shift Y left until Y >= X. */ |
| 82 |
l.ori r6, r0, 1 /* mask = 1 */ |
| 83 |
1: l.sflts r4, r0 /* y has msb set */ |
| 84 |
l.bf 2f |
| 85 |
l.sfltu r4, r12 /* y < x */ |
| 86 |
l.add r4, r4, r4 /* y <<= 1 */ |
| 87 |
l.bf 1b |
| 88 |
l.add r6, r6, r6 /* mask <<= 1 */ |
| 89 |
|
| 90 |
/* Shift Y back to the right again, subtracting from X. */ |
| 91 |
2: l.add r7, r11, r6 /* tmp1 = quot + mask */ |
| 92 |
3: l.srli r6, r6, 1 /* mask >>= 1 */ |
| 93 |
l.sub r8, r12, r4 /* tmp2 = x - y */ |
| 94 |
l.sfleu r4, r12 /* y <= x */ |
| 95 |
l.srli r4, r4, 1 /* y >>= 1 */ |
| 96 |
#if defined(__or1k_cmov__) |
| 97 |
l.cmov r11, r7, r11 /* if (y <= x) quot = tmp1 */ |
| 98 |
l.cmov r12, r8, r12 /* if (y <= x) x = tmp2 */ |
| 99 |
#else |
| 100 |
l.bnf 4f |
| 101 |
l.nop |
| 102 |
l.ori r11, r7, 0 |
| 103 |
l.ori r12, r8, 0 |
| 104 |
4: |
| 105 |
#endif |
| 106 |
l.sfne r6, r0 /* loop until mask == 0 */ |
| 107 |
l.bf 3b |
| 108 |
l.add r7, r11, r6 /* delay fill from loop start */ |
| 109 |
|
| 110 |
9: l.jr r9 |
| 111 |
l.nop |
| 112 |
|
| 113 |
.size __udivsi3, . - __udivsi3 |
| 114 |
.size __udivmodsi3_internal, . - __udivmodsi3_internal |
| 115 |
#endif |
| 116 |
|
| 117 |
#ifdef L__umodsi3 |
| 118 |
.balign 4 |
| 119 |
.global __umodsi3 |
| 120 |
.type __umodsi3, @function |
| 121 |
.cfi_startproc |
| 122 |
__umodsi3: |
| 123 |
/* Know that __udivmodsi3_internal does not clobber r13. */ |
| 124 |
l.ori r13, r9, 0 |
| 125 |
.cfi_register 9, 13 |
| 126 |
l.jal __udivmodsi3_internal |
| 127 |
l.nop |
| 128 |
l.jr r13 /* return to saved lr */ |
| 129 |
l.ori r11, r12, 0 /* move remainder to rv */ |
| 130 |
|
| 131 |
.cfi_endproc |
| 132 |
.size __umodsi3, . - __umodsi3 |
| 133 |
#endif |
| 134 |
|
| 135 |
/* For signed division we do: |
| 136 |
|
| 137 |
-x / y = x / -y = -(x / y) |
| 138 |
-x % y = -(x % y) |
| 139 |
x % -y = x % y |
| 140 |
|
| 141 |
which has the property that (x/y)*y + (x%y) = x. */ |
| 142 |
|
| 143 |
#ifdef L__divsi3 |
| 144 |
.balign 4 |
| 145 |
.global __divsi3 |
| 146 |
.type __divsi3, @function |
| 147 |
.cfi_startproc |
| 148 |
__divsi3: |
| 149 |
l.xor r6, r3, r4 /* need result negate? */ |
| 150 |
|
| 151 |
l.sflts r3, r0 /* abs(x) */ |
| 152 |
#if defined(__or1k_cmov__) |
| 153 |
l.sub r5, r0, r3 |
| 154 |
l.cmov r3, r5, r3 |
| 155 |
#else |
| 156 |
l.bnf 1f |
| 157 |
l.sub r5, r0, r3 |
| 158 |
l.ori r3, r5, 0 |
| 159 |
1: |
| 160 |
#endif |
| 161 |
l.sflts r4, r0 /* abs(y) */ |
| 162 |
#if defined(__or1k_cmov__) |
| 163 |
l.sub r5, r0, r4 |
| 164 |
l.cmov r4, r5, r4 |
| 165 |
#else |
| 166 |
l.bnf 2f |
| 167 |
l.sub r5, r0, r4 |
| 168 |
l.ori r4, r5, 0 |
| 169 |
2: |
| 170 |
#endif |
| 171 |
|
| 172 |
/* If the result will not require sign flip, tail call. */ |
| 173 |
l.sflts r6, r0 |
| 174 |
l.bnf __udivmodsi3_internal |
| 175 |
l.ori r13, r9, 0 /* save lr */ |
| 176 |
|
| 177 |
/* Otherwise, know that __udivmodsi3_internal does not clobber r13. |
| 178 |
Perform a normal call, then negate and return via saved lr. */ |
| 179 |
.cfi_register 9, 13 |
| 180 |
l.jal __udivmodsi3_internal |
| 181 |
l.nop |
| 182 |
l.jr r13 |
| 183 |
l.sub r11, r0, r11 |
| 184 |
|
| 185 |
.cfi_endproc |
| 186 |
.size __divsi3, . - __divsi3 |
| 187 |
#endif |
| 188 |
|
| 189 |
#ifdef L__modsi3 |
| 190 |
.balign 4 |
| 191 |
.global __modsi3 |
| 192 |
.type __modsi3, @function |
| 193 |
.cfi_startproc |
| 194 |
__modsi3: |
| 195 |
l.sflts r4, r0 /* abs(y) */ |
| 196 |
#if defined(__or1k_cmov__) |
| 197 |
l.sub r5, r0, r4 |
| 198 |
l.cmov r4, r5, r4 |
| 199 |
#else |
| 200 |
l.bnf 2f |
| 201 |
l.sub r5, r0, r4 |
| 202 |
l.ori r4, r5, 0 |
| 203 |
2: |
| 204 |
#endif |
| 205 |
|
| 206 |
l.sflts r3, r0 /* x negative? */ |
| 207 |
l.bf 1f |
| 208 |
l.ori r13, r9, 0 /* save lr */ |
| 209 |
|
| 210 |
/* Know that __udivmodsi3_internal does not clobber r13. */ |
| 211 |
.cfi_register 9, 13 |
| 212 |
|
| 213 |
/* X positive; no negate of the result required. */ |
| 214 |
l.jal __udivmodsi3_internal |
| 215 |
l.nop |
| 216 |
l.jr r13 /* return to saved lr */ |
| 217 |
l.ori r11, r12, 0 /* move remainder to rv */ |
| 218 |
|
| 219 |
/* X negative; negate both X and the result. */ |
| 220 |
1: l.jal __udivmodsi3_internal |
| 221 |
l.sub r3, r0, r3 |
| 222 |
l.jr r13 /* return to saved lr */ |
| 223 |
l.sub r11, r0, r12 /* negate remainder to rv */ |
| 224 |
|
| 225 |
.cfi_endproc |
| 226 |
.size __modsi3, .- __modsi3 |
| 227 |
#endif |